From b75e8f5fe395193e368b91aa615080f79a82ef43 Mon Sep 17 00:00:00 2001 From: Kaveen Kumarasinghe Date: Sat, 13 Jan 2024 04:20:00 -0500 Subject: [PATCH] Update PERMANENT-MEMORY.md Signed-off-by: Kaveen Kumarasinghe --- detailed_guides/PERMANENT-MEMORY.md | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/detailed_guides/PERMANENT-MEMORY.md b/detailed_guides/PERMANENT-MEMORY.md index f7786d6b..6bb3af99 100644 --- a/detailed_guides/PERMANENT-MEMORY.md +++ b/detailed_guides/PERMANENT-MEMORY.md @@ -1,5 +1,7 @@ -# Permanent Memory and Conversations -Permanent memory has now been implemented into the bot, using the OpenAI Ada embeddings endpoint, and Pinecone. +# Permanent Memory and Conversations +We are migrating towards using [QDRANT](https://qdrant.tech/) as our vector database backing, we are moving away from pinecone. Qdrant is an excellent vector database choice, and in fact the best one that we've tested and used so far. + +Permanent memory has now been implemented into the bot, using the OpenAI Ada embeddings endpoint, and Pinecone. Pinecone is a vector database. The OpenAI Ada embeddings endpoint turns pieces of text into embeddings. The way that this feature works is by embedding the user prompts and the GPT responses, storing them in a pinecone index, and then retrieving the most relevant bits of conversation whenever a new user prompt is given in a conversation. @@ -22,4 +24,4 @@ To manually create an index instead of the bot automatically doing it, go to the Then, name the index `conversation-embeddings`, set the dimensions to `1536`, and set the metric to `DotProduct`: -
\ No newline at end of file +